Research Computing’s Fortress storage archive expands capacity
The Fortress archival storage system has a new tape library that has expanded the archive’s capacity by a factor of five, from approximately 50 petabytes of data to more than 250 petabytes.
The new SpectraLogic TFinity tape library can hold 9,000 tapes, of a new generation that can hold about four times as much data as the previous generation. They are also twice as fast, meaning data can be read off the tapes at 600 megabytes per second instead of the previous 300.
The new tape library has two arms, which means it can handle more requests at the same time, significantly reducing the mean time between when a user requests data and when it’s available.
“This upgrade is phase one of a multi-phase improvement to the archive system that will include more disk capacity, faster networking and newer, faster servers to handle the ever-growing demands of researchers looking for a safe place to hold their data long term,” says Ramon Williamson, senior engineer for Research Computing. “Tape continues to be the most cost effective way to deliver large-scale archival storage, and these improvements will help make the archiving experience faster, easier and more reliable for end-users.”
Researchers in fields such as climate modeling, agriculture, physics, chemical modeling and life sciences have long depended on Fortress for a place to store their growing datasets from increasingly more complicated models.
This has led to an exponential growth in the amount of data being ingested into the archive. From first reaching the one petabyte mark in 2013, the archive has grown to its current 60 petabytes of stored data in nine years.
“The Fortress archive is a key component of the set of tools that we provide to researchers to manage the research data lifecycle. High-speed scratch on community cluster supercomputers, the Data Depot and Fortress all have a distinct role to play in managing the data collected from scientific instruments at Purdue, and produced as the output of modeling and simulation workflows”, says Preston Smith, Executive Director for Research Computing.
To learn more about Fortress or other Research Computing resources, contact firstname.lastname@example.org.
Writer: Adrienne Miller, science and technology writer, Research Computing, email@example.com.